The latest Pfizer safety report shows a massive cardiovascular signal
and also shows some very odd data.
Let's start with the first. 🧵
NHR and PHARMO are different European databases.
The total number of cardiovascular events of any type works out at a risk of
1 in 264 for NHR
and
1 in 362 for PHARMO
By contrast the claimed risk of severe covid was reduced by
1 in 2403 for NHR
and
1 in 5181 for PHARMO
Pfizer admit there is a signal in the PHARMO database but they have a plan to make it disappear...
"it could be possible to model some constructive bias analysis to correct this in the final analysis."
If you had a new set of aircraft and there was a report of a problem from one country - would you ignore it because the others had not reported a problem?
Would you fudge the data from the concerning country by modelling it to look like the other countries?
There are differences between these datasets.
People in the PHARMO dataset were healthier - they had a far lower incidence of death, for example.
Signals are easier to see in healthier datasets because they are less likely to be lost in the background noise.
You may wonder why the deaths are higher in the unvaccinated.
The well recognised healthy vaccinee effect means that those who are about to die reject the vaccine and end up inflating the mortality of the unvaccinated.
However, Pfizer did not do analysis having excluded this period.
They also did no analysis by age group except for myocarditis.
Because of low background rates in younger age groups signals would be more obvious if these were analysed separately.
One way to reduce the risk of this bias would be to only look at people who were sick in the week before vaccination or matching (if unvaccinated).
This was the plan (although maybe they meant it as an exclusion criteria?):
Until the EMA stopped it.
They did not ask for analysis only of those who had contact with health care system prior to injection or matching!
Now for the odd data.
Given the healthy vaccinee effect, the unvaccinated should have higher rates of any condition in the first 60 days.
But they do not.
What is more odd is that the number of people is huge at the beginning but rapidly reduces.
This is because the unvaccinated are matched to the vaccinated. When they get vaccinated that pairing is dropped.
That should mean huge numbers of events in the first 60 days. e.g.
Independent evidence is consistent in showing a higher rate of covid in the first two weeks.
Pfizer claims the opposite (except for PHARMO)!
Despite the clear issues there are still really concerning signals like this one:
I have taken a deep dive to understand exactly what's happening with deaths of under 1 year olds.
Who wants to look at baby deaths?
I get it.
But DO NOT LOOK AWAY.
It is....
First of all I did this because of frustration with people arguing over what "expected" deaths should look like.
You can make up reasons for picking particular years and come up with a totally different story.
'99-'19 excess deaths from '21
'11-'16 deficit in deaths from '21
What can we do to see if the rise is meaningful?
First we can look monthly ('24 and '25 data is incomplete)
Here are the monthly deaths for females which rises from March 2021 (having been below expected before) and stays high except for deficit in winter 2021-2022.
All the vaccines ever did was make people immune suppressed for two weeks.
The consequence was that those who were susceptible to a particular variant had their infections earlier than they would have.
You can measure after that point and get an illusion of benefit.
This trick only works if you vaccinate during a wave.
For the UK - we were fast getting to those most susceptible to dying. Our death wave looks like a witch's hat on top of Europe's because of the earlier infections.